W2-Model Serving Architecture

Question 1

Two main places to run infrastructure for machine learning models?

Model Serving Architecture- GPT Summary

Answer 1

on-premises or outsourcing to a cloud provider.

Question 2

There are two main ways of scaling: vertical and horizontal. Vertical scaling is using ____ , whereas horizontal scaling is ____.

Scaling Infrastructure- GPT summary

Answer 2

bigger and more powerful hardware
adding more devices to the network

Question 3

Horizontal scaling is generally recommended, why?

Scaling Infrastructure- GPT summary

Answer 3

Because it provides elasticity and avoids taking the app offline to upgrade hardware resources. Moreover, it can be done automatically by the system, which reduces overall latency.